Note on scope: The workshop participant group was largely drawn from people already working in or closely adjacent to the CM field, and as
Bert Frohlich noted in S3, "we are kind of the believers." The cruxes and questions below reflect the state of disagreement
within that group. External skeptics — including researchers who have published pessimistic analyses — were not well-represented in the live session. David Reinstein noted in S3 that he intends to engage those voices directly. The "key questions for skeptics" section below represents what optimistic workshop participants would want skeptics to address; it is not a set of answers.
Before mapping disagreements, it helps to note what workshop participants broadly converged on. These are not fully settled — but they represent positions where the workshop did not surface active disagreement.
- Humbird's amino acid cost assumptions are overstated. GFI's December 2025 analysis obtained real bulk supplier quotes and found current purchasable prices 2–10× below Humbird's projections. This is an empirical update, not a modeling assumption, and was not challenged by any participant.
- A transition from pharma-grade to food-grade inputs is achievable and likely. Bomkamp: "The switch to food-grade ingredients represents a very substantial 'low-hanging fruit' cost reduction opportunity — I would therefore expect that few products beyond the very earliest launched would use substantial amounts of pharma-grade inputs." Multiple respondents gave 100% probability (CM_17) to this before 2036.
- Precision fermentation and plant molecular farming will reduce growth factor costs substantially. Fuchs gives 100% probability; Swartz agrees the direction is clear, though he's more cautious about timeline. No participant argued GF costs are permanently high.
- CM's animal welfare case is primarily a timeline and probability question, not a feasibility question. Almost all participants believe CM can succeed at scale. The debate is about when, at what cost, and with what probability — not whether it is physically possible.
- Density alone is a misleading performance metric. Cell volume per milliliter, performance-to-cost ratio (Frohlich/Swartz), and protein/lipid content (Swartz, McNulty) are more useful. There is not yet a shared standard metric for comparing bioprocess performance across scenarios.
- The field has critical data gaps that make model validation impossible right now. Swartz (S3 transcript): "There's basically no published studies that really quantify feed conversion ratio — what it actually is. There's also no information about inclusion rates... very little information around what is the actual cost of equipment."
A crux is a specific empirical or technical question where: (a) different answers lead to meaningfully different 2036 cost projections; (b) reasonable, well-informed people currently disagree; and (c) the question is in principle resolvable by evidence. The five cruxes below are ordered roughly from most-upstream to most-downstream in the production process.
Crux 1
Cell line engineering: performance cascade, not just growth factor independence
Cell line strategy is the most upstream cost-relevant decision in CM production. As Kubinyecz (S1) put it: "Immortalisation strategy determines proliferation rates, achievable density, growth factor dependence, and media requirements." Gene editing is one tool; the crux is not specifically whether gene editing succeeds, but what level of cell line performance — across all these dimensions — is commercially validated by 2030–2036.
Sub-crux 1a — GF independence via gene editing: Will autocrine-engineered cell lines (cells that produce their own growth factors) reach commercial-scale validation? Fuchs estimates 20% probability by 2036; Swartz similarly cautious. Most published TEAs do not model this pathway as a primary scenario. If it succeeds, the GF cost component collapses. If it doesn't, the field depends on external GF reduction pathways.
Sub-crux 1b — GF cost via non-gene-editing pathways: Precision fermentation and plant molecular farming as sources of cheap recombinant GFs. Fuchs: 100% probability by 2036. Swartz: directionally correct, timeline uncertain. Small molecule substitutes and thermostable variants are additional pathways discussed; both have remaining scientific uncertainty but are not ruled out.
Sub-crux 1c — Regulatory geography: US FSMA has cleared gene-edited cell lines for most food applications. EU posture is more conservative, with EFSA scrutiny analogous to GM food. Italy's 2023 production ban is a concrete data point. Where commercial scale first happens may therefore depend on regulatory jurisdiction — which affects the timeline and economics for most participants.
Key sources: Kubinyecz S1 framing; Fuchs CM_12/CM_13 beliefs form; Swartz CM_12 written comment; S3 transcript (GF pathway discussion)
Crux 2
Media: hydrolysate substitution and the limits of media cost reduction
Two distinct steps in media cost reduction are often conflated. Workshop participants were broadly aligned on the first and actively divided on the second.
Step 1 (broad agreement): pharma-grade → food-grade inputs. Bomkamp describes this as "low-hanging fruit" that should happen as regulatory frameworks normalise. Multiple participants gave 100% probability to this transition before 2036 (CM_17 in beliefs form). This step reduces costs substantially but does not eliminate them.
Step 2 (active crux): replacing purified amino acids with hydrolysates. Hydrolysates — plant-derived or by-product protein digests — cost orders of magnitude less than purified AAs at commodity scale. Fuchs (ACIB/FEASTS) argues full or near-full substitution is achievable. Swartz and Bomkamp say partial substitution (~80%+) is plausible but full replacement faces unresolved QC barriers. Swartz's CM_12 comment: "Hydrolysates may become incorporated in a decent number of manufacturers' media formulations over the next decade. It is unlikely you'd fully replace purified AAs. You'd still likely need to supplement them in. It's also unclear if hydrolysates are truly cost beneficial."
The key unresolved question, as Swartz put it at the end of S1: "The biggest question for me is whether the hydrolysates can meet nutritional needs for high-density growth in suspension settings, and whether this is actually more economically efficient than using purified nutrients." The cost difference between a Humbird-type scenario (largely purified AAs) and a Fuchs-type scenario (near-full hydrolysate substitution) is substantial — but the specific disagreement between workshop participants is about whether full substitution is achievable and net cost-positive at commercial density, not about whether some substitution will happen.
Would be very interesting to know if feed/energy conversion is improved by small peptide uptake. That would be a strong justification for using hydrolysates that I don't see discussed much.
— Elliot Swartz, S1 chat
Key sources: Fuchs S1 presentation; Swartz CM_12 written comment; Bomkamp CM_17; S1 chat log; S1 transcript discussion
Crux 3
Cell productivity: density, process mode, FCR, and the binding constraint
Even if media costs fall substantially, the remaining hard problems are cell productivity (density × growth rate × bioreactor occupancy) and capital costs. Swartz's pre-workshop submission: "Media costs will be quite low. Questions remain around productivity and capital costs."
Cell density: Optimistic TEAs model 20–50 g/L. More conservative analyses work from lower figures consistent with current practice. A 2× gap in density assumption propagates as 2× more media cost per kg output. Tarka Abraham (Ivy Farm, S1) is an industry practitioner working at commercial-scale densities — their perspective on achievable density in 20kL+ bioreactors was flagged as among the most grounded real-world data points.
Process mode — fed-batch vs. perfusion: Frohlich (S3): "The economics of continuous processes, and in this case perfusion, I think are pretty undeniable if you do the numbers." But CDMO infrastructure is largely fed-batch-based, and contamination risk at CM-scale perfusion (two orders of magnitude beyond biopharma) is real and not yet resolved. Suspension vs. adherent for differentiated products was also flagged as unresolved.
Feed conversion ratio: No published studies quantify this directly for CM cell lines (Swartz, S3 transcript). This is a named data gap — not an area of disagreement, but an area of unknown that prevents model validation.
Performance-to-cost ratio: Frohlich and Swartz's November 2025 paper proposed this as a better organizing metric: how much biomass can a given capital investment produce? This is independent of media costs and captures the capital × productivity interaction that density alone misses.
There's basically no published studies that really quantify feed conversion ratio — what it actually is. There's also no information about inclusion rates — what inclusion rates are actually going to yield tasty products. There's also very little information around what is the actual cost of equipment. How much capital do you need to build a facility of a certain size? Is a scale-up or scale-out approach more prudent, given capital constraints?
— Elliot Swartz, S3 transcript
Key sources: Swartz S3 transcript (data gaps); Frohlich S3 transcript (process mode, performance-to-cost ratio); Abraham S1 (density, bioreactor scale); S3 chat (density, metrics discussion)
Crux 4
The product target: what is the end-point, and how does it constrain cost?
This crux was not in the pre-workshop framing but emerged prominently in S3. It may be more fundamental than the cost cruxes above: if the product target is undefined or wrong, optimizing for biomass cost is optimizing the wrong thing.
The product-definition gap (McNulty, Frohlich): Most current CM R&D proceeds from a biomass-production perspective rather than reverse-engineering from a target food product. Frohlich (S3): "I haven't seen a good analysis of what would make a meat substitute that is perhaps a hybrid of using plant and mycoprotein along with animal cells. How many animal cells would you actually need in the product to give it that flavor?" McNulty (S3): "We don't understand specifically what are the product attributes that we need to have at the end point to have a viable product that's going to drive market adoption."
Inclusion rate as a cost lever: If a hybrid product (CM cells + plant protein or mycoprotein) at 5–25% inclusion can deliver sensory parity at a fraction of the cost of a 100% CM product, the cost-parity timeline changes substantially. Bomkamp (S3) cited a 2025 paper where 1.2% heat-treated cell extract produced a substantial sensory boost in a hybrid Gyoza product. Frohlich argues this direction is underexplored: "Can we make a hybrid product... and get a reasonable taste? And if an investor were able to see and taste something like that, I think the money would be forthcoming."
"Price parity" requires a target (Bomkamp): Price parity with a Michelin-starred beef product is a near-term possibility. Price parity with commodity ground beef is a long-term target. Framing CM cost progress as a single threshold misrepresents the actual market entry pathway, which Bomkamp describes as "more of a journey than a single destination."
You've got to start with the end in mind. A lot of the companies I've been consulting for, it's like, well, let's make this biomass slop and then figure out what to do with it. It should be the other way around: what are you trying to make, and therefore design the cell culture process to provide the ingredient that's of importance to the final product?
— Bert Frohlich, S3 transcript
Key sources: Frohlich S3 transcript; McNulty S3 transcript; Bomkamp S3 transcript (inclusion rate, price parity framing)
Crux 5
Capital availability and industry survival: will the field reach dedicated-plant scale?
This crux is different in kind from Cruxes 1–4. Those concern what 2036 production costs will be if the industry reaches commercial scale. This one concerns whether it does. The workshop organizer noted in S3 that CDMO infrastructure is not the right focus for forecasting 2036 costs — "that's an intermediate testing ground" — but the capital question is real and unresolved.
The funding gap: Swartz (S3): "What the industry needs is money that is gonna yield steel in the ground, and that is the hardest thing to come by right now." GFI's GINA framework estimates $10 billion/year of public + private investment is needed for alternative proteins broadly. Current public funding is far below that. Private funding has contracted significantly since 2022.
The CDMO-to-dedicated-plant transition: CDMOs provide the best current empirical data on production costs. But TEAs model a hypothetical at-scale dedicated facility, not CDMO economics. The transition requires tens to hundreds of millions per facility — capital most current CM companies cannot access. This timeline affects whether commercial-scale dedicated plants exist in any meaningful quantity by 2036.
Regulatory survival risk: Italy's 2023 CM production ban directly affected Bruno Cell (Lattanzi). EU regulatory timelines are long. If early commercial scale is US-only or Singapore-only, the global trajectory is different from a scenario where EU and UK markets open by 2030.
Note on the CM_01 question: Crux 5 does not directly change what it costs to produce CM biomass at a hypothetical 2036 dedicated facility — that's what CM_01 asks. But it does affect the probability distribution over whether any such facility exists. If the industry does not survive the funding drought, the "2036 cost" becomes academic. This is why Lattanzi's pessimistic cost estimate ($100/kg) does not translate to pessimism about the value of CM investment — he assigns high value to CM if it succeeds, and his risk assessment is primarily about timeline and capital, not ultimate technical feasibility.
Key sources: Swartz S3 transcript (capital, market shaping); Frohlich S3 transcript (government investment, GINA); Lattanzi beliefs form CM_02; S2 (off-record; only high-level themes from S2 are cited here)
In S3, David Reinstein asked: "What would we want to put forward to the skeptics? What would you want them to answer? What am I getting wrong here?" Bert Frohlich framed it as: "What would you want to see that would convince you that this is a reality?" The questions below are derived from that discussion — specifically, they are claims the workshop's optimistic participants make that a well-informed skeptic should engage with, rather than dismiss.
Q1 · Media costs
GFI's December 2025 analysis found real bulk supplier quotes for key amino acids 2–10× below Humbird's projections. Given that, what is your revised estimate for basal media cost at food-grade, industrial scale in 2036 — and what specific input or assumption prevents it from falling further?
What would update this: detailed bulk supplier data by amino acid type; experimental evidence on whether hydrolysates meet nutritional needs at commercial suspension density.
Q2 · Cell line performance trajectory
CHO cells in biopharma went from ~1–2 million cells/mL (1980s) to 100+ million/mL under perfusion, over 40 years of focused optimization. CM-specific cell lines have had ~5–10 years. At what rate do you expect CM cell line performance to improve on the relevant metrics (density, FCR, GF independence), and why is that rate different from CHO's trajectory? What specific biological or engineering constraint creates the ceiling?
What would update this: published FCR data for CM cell lines; density benchmarks from companies willing to share (even non-attributed); results from open academic programs (NICA, ACIB, FEASTS).
Q3 · Growth factor cost pathways
Fuchs estimates 100% probability that precision fermentation and plant molecular farming will produce recombinant growth factors at cost-competitive prices by 2036. Swartz sees this as directionally correct. What is your probability estimate, and what specific technical or commercial obstacle prevents this from happening within a decade in at least one jurisdiction?
What would update this: demonstration at industrial fermentation scale; regulatory clearance for recombinant GF use in CM production in at least one major jurisdiction.
Q4 · Inclusion rates and hybrid products
If a hybrid CM product (animal cells + plant protein or mycoprotein) at 5–20% animal cell inclusion can deliver sensory parity with conventional meat in relevant market segments, the cost-parity threshold is 5–20× lower than if 100% animal cells are required. What is your estimate of the minimum effective inclusion rate for a commercially viable CM product, and on what evidence?
What would update this: published sensory studies with varied inclusion rates; investor willingness-to-fund data on hybrid products; regulatory clarity on labelling requirements for hybrid products.
Q5 · The learning-curve analogy
Solar PV and battery storage both saw 80–90% cost reductions over roughly a decade once manufacturing scaled up, driven by learning curves, R&D, and competitive pressure. Swartz suggests a similar trajectory is plausible for CM given enough capital investment. What makes CM's cost trajectory more like biopharma (costs plateau due to biological and regulatory constraints) than solar/batteries (steep learning curve with scale)? Is this a claim about biology, about regulation, or about market structure?
What would update this: evidence of learning-curve effects in early production runs; successful demonstration of cost reduction from pilot to pilot-scale within a single company.
Q6 · Regulatory and geographic scope
US FSMA has cleared gene-edited cell lines for most food applications. Singapore has an established novel food regulatory pathway. If early commercial scale is US and/or Singapore only for 2026–2032 (i.e., grant that EU regulatory constraints bind for a decade), what does the cost trajectory look like, and does that change your 2036 estimate for the global cost frontier?
What would update this: commercial approvals in multiple jurisdictions; demonstrated commercial production at scale in a permissive jurisdiction; EU EFSA decisions on gene editing in cultivated animal products.
Key questions for optimists — forthcoming. The questions above represent what workshop participants (a largely optimism-leaning group) want skeptics to engage with. A symmetric set — what data or outcomes would change the minds of optimists, and what would constitute evidence that scale-up is not going to work — will be added once input from researchers who hold explicitly skeptical positions has been gathered. Claire Bomkamp (S3): "If something came out that was like, hey, this is going to be an insurmountable obstacle on sort of any of these cost parameters, I think any of those things independently could change my mind... The stuff that is more likely to come up that would change my mind would be more on the bioprocess side."
The crux map above suggests several places where the interactive cost model dashboard needs development or explicit scenario treatment:
Cell line scenario branching: The model should represent two distinct scenarios — one with optimized, potentially gene-edited cell lines achieving high density and GF independence by 2033, and one without. The cost gap between these scenarios (Crux 1) dwarfs any single parameter change elsewhere in the model.
Hydrolysate substitution as a slider, not a binary: Rather than assuming full hydrolysate substitution (Fuchs) or minimal substitution (Humbird), the model should parameterize substitution rate (0–100%) alongside a QC cost factor that increases with substitution rate. This makes the Swartz/Fuchs crux explicit in the model output (Crux 2).
FCR as a named unknown: The model currently does not represent feed conversion ratio explicitly. Given Swartz's statement that no published data quantifies this, it should be represented as a wide prior rather than a fixed assumption, with sensitivity analysis showing how cost changes as FCR varies (Crux 3).
Inclusion rate as a separate cost lever: Bomkamp's work on translating biomass cost to product cost via inclusion rate is not currently in the model. A simple inclusion-rate slider (1%–100%) showing final product cost would make Crux 4 legible to model users and funders.
Capital availability is a path variable, not a cost parameter: The CDMO-to-dedicated-plant transition and the capital availability question (Crux 5) cannot be represented as a cost input to the 2036 model — they affect the probability that the 2036 scenario occurs at all. This distinction should be explicit in how the model is presented.